olua commited on
Commit
b79e804
·
verified ·
1 Parent(s): 2129dc7

Upload tokenizer

Browse files
Files changed (3) hide show
  1. added_tokens.json +2 -2
  2. tokenizer_config.json +4 -4
  3. vocab.json +43 -46
added_tokens.json CHANGED
@@ -1,4 +1,4 @@
1
  {
2
- "</s>": 50,
3
- "<s>": 49
4
  }
 
1
  {
2
+ "</s>": 47,
3
+ "<s>": 46
4
  }
tokenizer_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "added_tokens_decoder": {
3
- "47": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
@@ -8,7 +8,7 @@
8
  "single_word": false,
9
  "special": false
10
  },
11
- "48": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
@@ -16,7 +16,7 @@
16
  "single_word": false,
17
  "special": false
18
  },
19
- "49": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
@@ -24,7 +24,7 @@
24
  "single_word": false,
25
  "special": true
26
  },
27
- "50": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
 
1
  {
2
  "added_tokens_decoder": {
3
+ "44": {
4
  "content": "[UNK]",
5
  "lstrip": true,
6
  "normalized": false,
 
8
  "single_word": false,
9
  "special": false
10
  },
11
+ "45": {
12
  "content": "[PAD]",
13
  "lstrip": true,
14
  "normalized": false,
 
16
  "single_word": false,
17
  "special": false
18
  },
19
+ "46": {
20
  "content": "<s>",
21
  "lstrip": false,
22
  "normalized": false,
 
24
  "single_word": false,
25
  "special": true
26
  },
27
+ "47": {
28
  "content": "</s>",
29
  "lstrip": false,
30
  "normalized": false,
vocab.json CHANGED
@@ -1,51 +1,48 @@
1
  {
2
  "\n": 0,
3
  "#": 2,
4
- "$": 3,
5
- "(": 4,
6
- ")": 5,
7
- "/": 6,
8
- "0": 7,
9
- "1": 8,
10
- "2": 9,
11
- "3": 10,
12
- "4": 11,
13
- "5": 12,
14
- "6": 13,
15
- "7": 14,
16
- "8": 15,
17
- "9": 16,
18
- "@": 17,
19
- "[PAD]": 48,
20
- "[UNK]": 47,
21
- "a": 18,
22
- "b": 19,
23
- "c": 20,
24
- "d": 21,
25
- "e": 22,
26
- "f": 23,
27
- "g": 24,
28
- "h": 25,
29
- "i": 26,
30
- "j": 27,
31
- "k": 28,
32
- "l": 29,
33
- "m": 30,
34
- "n": 31,
35
- "o": 32,
36
- "p": 33,
37
- "q": 34,
38
- "r": 35,
39
- "s": 36,
40
- "t": 37,
41
- "u": 38,
42
- "v": 39,
43
- "w": 40,
44
- "x": 41,
45
- "y": 42,
46
- "z": 43,
47
  "|": 1,
48
- "": 44,
49
- "": 45,
50
- "’": 46
51
  }
 
1
  {
2
  "\n": 0,
3
  "#": 2,
4
+ "(": 3,
5
+ ")": 4,
6
+ "/": 5,
7
+ "0": 6,
8
+ "1": 7,
9
+ "2": 8,
10
+ "3": 9,
11
+ "4": 10,
12
+ "5": 11,
13
+ "6": 12,
14
+ "7": 13,
15
+ "8": 14,
16
+ "9": 15,
17
+ "[PAD]": 45,
18
+ "[UNK]": 44,
19
+ "a": 16,
20
+ "b": 17,
21
+ "c": 18,
22
+ "d": 19,
23
+ "e": 20,
24
+ "f": 21,
25
+ "g": 22,
26
+ "h": 23,
27
+ "i": 24,
28
+ "j": 25,
29
+ "k": 26,
30
+ "l": 27,
31
+ "m": 28,
32
+ "n": 29,
33
+ "o": 30,
34
+ "p": 31,
35
+ "q": 32,
36
+ "r": 33,
37
+ "s": 34,
38
+ "t": 35,
39
+ "u": 36,
40
+ "v": 37,
41
+ "w": 38,
42
+ "x": 39,
43
+ "y": 40,
44
+ "z": 41,
 
 
45
  "|": 1,
46
+ "": 42,
47
+ "": 43
 
48
  }